Chapter 28: The Professor
Professor Kim Jongwoo’s office was on the fourth floor of the SNU Engineering building, behind a door that was always slightly ajar and covered in Post-it notes written in a handwriting that looked like a seismograph during an earthquake.
The notes said things like “MEETING CANCELLED (again)” and “DO NOT DISTURB UNLESS YOU HAVE RESULTS” and, inexplicably, “BRING COFFEE IF ENTERING.” Daniel had been reading these notes from the hallway for three weeks, attending Professor Kim’s lectures from the back row of a class he wasn’t enrolled in, and waiting for the right moment to introduce himself.
The right moment arrived on a Thursday afternoon in May, when Professor Kim’s lecture on natural language processing was interrupted by a student’s question that was so fundamentally wrong that the professor actually stopped mid-sentence, removed his glasses, and stared at the ceiling for eight full seconds before responding.
“Mr. Park,” Professor Kim said, his voice carrying the weary patience of a man who had been explaining the same concepts for thirty years and had come to accept that most people would never understand them, “neural networks do not ‘think.’ They process. There is a difference so profound that confusing the two is like confusing the map with the territory.”
“But Professor, the papers say—”
“The papers say what the papers say. The papers are written by humans who use metaphors because the truth is too complex for most audiences. The truth is that a neural network is a mathematical function. An extremely sophisticated mathematical function, but a function nonetheless. It does not have consciousness, intention, or opinions about the weather.”
A hand went up in the back row. Daniel’s hand. He hadn’t planned to raise it. It just happened—the instinct of a man who had spent twenty years in rooms where the wrong understanding of AI could cost billions of dollars.
Professor Kim’s eyes found him. Squinted. “You’re not in this class.”
“No, sir. I’m auditing.”
“Business Administration?”
“Yes, sir.”
“Business students don’t audit NLP courses.” Professor Kim put his glasses back on. “But you have your hand up, which means you either have a question or you think you have an answer. Which is it?”
“An observation, sir. You said neural networks are mathematical functions. That’s true in the current architecture. But what if the architecture itself evolves? If you layer enough transformer blocks with self-attention mechanisms, you get emergent behaviors that can’t be predicted from the individual components. The function becomes more than the sum of its parts.”
The lecture hall went very quiet.
Professor Kim removed his glasses again. This time, it wasn’t the weary gesture of a man dealing with ignorance. It was the sharp, focused gesture of a man who had just heard something unexpected from a direction he wasn’t watching.
“Transformer architectures,” he said slowly. “Self-attention mechanisms. You’re describing a concept that my research team has been working on for two years. It hasn’t been published. How do you know about it?”
Because in 2017, a paper called “Attention Is All You Need” will revolutionize AI. Because transformer architectures will become the foundation of GPT, BERT, and every large language model that reshapes the world. Because I’ve read that paper approximately fifty times in my first life, and you, Professor Kim, are one of the people whose work laid the groundwork for it.
“I read your 2007 paper on sequence-to-sequence models,” Daniel said carefully. “The one in the Korean Journal of Computer Science. The attention concept you described in section 4.3 seemed like a natural extension of the transformer idea. I extrapolated.”
“You extrapolated.” Professor Kim’s voice was flat, but his eyes were sharp. “A business freshman extrapolated from a technical paper to a concept that most PhD students in my department haven’t grasped.”
“I’m a fast reader.”
“Being a fast reader doesn’t explain understanding.” Professor Kim looked at his watch. “Class dismissed. Mr.—”
“Cho. Daniel Cho.”
“Mr. Cho. My office. Now.”
Professor Kim’s office was exactly what Daniel remembered from his first life: organized chaos. Every surface was covered in papers, journals, and printouts of code. A whiteboard took up one entire wall, filled with equations and diagrams connected by arrows that followed a logic only the professor understood. On the desk, between stacks of student papers, was a half-eaten kimbap that had been there long enough to suggest a complicated relationship with the concept of lunch.
“Sit,” the professor said, pointing to the one chair that wasn’t covered in papers. He sat behind his desk, removed his glasses, cleaned them on his shirt, and put them back on—a ritual that Daniel recognized as the professor’s way of buying time to think.
“How old are you?” Professor Kim asked.
“Nineteen.”
“And you’re studying business.”
“Yes.”
“Why business? With your understanding of NLP, you could be in computer science.”
“Because I want to build a company that commercializes the kind of research you’re doing. Not in five years or ten years. Now. The technology is closer to practical application than most people realize.”
Professor Kim leaned back in his chair. It creaked—an old chair, the kind that universities provided to professors as a reminder that comfort was not a priority. “And you think you know when the technology will be ready?”
“I think the transformer architecture you’re developing—the self-attention mechanism—will be the foundation of the next revolution in AI. Not symbolic AI, not traditional machine learning. Deep learning with attention. It will change everything.”
“That’s a bold claim from a nineteen-year-old.”
“I’ve been told I’m bold.”
“Bold is a polite word for something that could also be called reckless.” But the professor was leaning forward now, his lunch forgotten, his glasses slightly askew. “Tell me specifically what you think transformer architectures will be used for.”
“Language understanding. Translation. Text generation. Eventually, multimodal processing—combining text, images, and audio in a single model. The applications for business are enormous: customer service automation, content generation, data analysis, code generation—”
“Code generation?” Professor Kim’s eyebrows rose. “You think a neural network will write code?”
“I think a sufficiently large language model, trained on enough code examples, will be able to generate functional code from natural language descriptions. Not perfectly. Not reliably. But well enough to augment human programmers.”
By 2023, GitHub Copilot will be doing exactly this, and it will change the software industry forever. But I can’t tell you that, because it hasn’t happened yet.
The professor was quiet for a long time. The office hummed with the sound of an old computer fan and the distant murmur of students in the hallway. Through the window, the SNU campus was green with May—trees in full leaf, students on the grass, the specific beauty of a place where young people pretended to study while actually talking about everything else.
“Mr. Cho,” Professor Kim said. “I’ve been in academia for thirty years. I’ve had exactly three students who genuinely understood the implications of my research. One is now at Google. One is at MIT. The third dropped out to become a monk, which I respect but find wasteful.”
“And me?”
“You’re not a student of mine. You’re a business major who wandered into my lecture and described my unpublished research with the confidence of someone who’s seen the future.” He paused. “That either makes you very smart or very well-connected. Which is it?”
“Can it be both?”
“It can be whatever you want, as long as you’re honest about your intentions.” Professor Kim stood and walked to the whiteboard. He picked up a marker and drew a circle. “This is where AI is now.” He drew a larger circle around it. “This is where it could be in ten years. The gap between these two circles is what I’ve spent my career trying to close.”
“And you’re doing it alone.”
“I’m doing it with a research team of six graduate students, a budget that wouldn’t cover the catering at a Samsung shareholder meeting, and a university administration that thinks AI is a fad.” He capped the marker. “So when a nineteen-year-old business student walks into my office and talks about commercializing my research, I’m interested. But I’m also cautious.”
“What would make you less cautious?”
“Time. Results. And the understanding that my research is not a product to be packaged and sold. It’s knowledge. Knowledge should be free.”
“I agree. The knowledge should be free. The applications built on top of that knowledge—those can be products. And the revenue from those products can fund more research. It’s a virtuous cycle.”
“A virtuous cycle.” Professor Kim smiled—the first smile Daniel had seen from him. It was a cautious smile, the smile of a man who wanted to believe but had been disappointed too many times. “You sound like a businessman.”
“I am a businessman. But I’m also a student. And I’m asking you to let me attend your seminars, read your papers, and learn from you. Not as a customer. As a student.”
“You want to learn NLP as a business major?”
“I want to understand the technology well enough to build a company around it. Not now. In two years. Maybe three. But when I’m ready, I want to come to you with a real proposal—not a pitch deck, not a dream, but a concrete plan with funding, a team, and a timeline.”
Professor Kim looked at him. The look was long, thorough, the kind of assessment that a man applies to a problem he hasn’t solved yet but suspects might be worth solving.
“My seminars are Thursday evenings, 6 to 8 PM, room 401,” he said. “You’ll be the only non-CS student. The others will eat you alive.”
“I’ll bring coffee.”
“The good kind. Not that vending machine poison.”
“Noted.”
“And Mr. Cho?”
“Yes?”
“If you’re going to sit in my lectures, at least enroll as an auditor. It irritates me when people sit in the back like ghosts.”
“I’ll file the paperwork today.”
“Good. Now leave. I have a kimbap to finish.”
Daniel left the office. In the hallway, the Post-it notes fluttered in the breeze from an open window. BRING COFFEE IF ENTERING.
Professor Kim Jongwoo. The third piece. In my first life, it took three months of careful cultivation to earn his trust. This time, it took a single lecture and an honest conversation.
Three for three. Sarah. Marcus. Kim.
The team is assembling. The pieces are falling into place.
But pieces on a board don’t make a game. For that, you need a plan. And a plan needs time.
He walked across the campus in the warm May evening, already composing the email he’d send to Sarah and Marcus: Tuesday coffee. I have someone you need to meet.
The company didn’t have a name yet. It didn’t have a product. It didn’t even have a formal agreement between its three founders-to-be. But it had something that was worth more than all of those things combined.
The right people, in the right room, at the right time.
Everything else was just execution.